A system and method implemented on a computing device for analyzing a digital corpus of unstructured interlocutor conversations to discover intents, goals, or both intents and goals of one or more parties to the conversations, by grouping the conversation utterances according to semantic similarity clusters; selecting the best utterance(s) that mostly likely embody a party's stated goal or intent; creates a set of candidate intent names for each cluster based upon each intent utterance in each conversation in each cluster; rates each candidate intent (or goal) for each intent name; and selects the most likely candidate intent (or goal) name for the purposes of subsequent automation of future conversations such as, but not limited to, automated electronic responses using Artificial Intelligence and machine learning.
Legal claims defining the scope of protection, as filed with the USPTO.
2. The method of claim 1 wherein the grouping is preceded by encoding sentence embeddings contained in the corpus.
3. The method of claim 2 wherein the encoding sentence embeddings comprises performing Language-Agnostic Bidirectional Encoder Representations from Transformers Sentence Encoding (LABSE).
4. The method of claim 2 wherein the encoding sentence embeddings comprises performing Robustly Optimized Bidirectional Encoder Representations from Transformers Pretraining Approach (RoBERTa).
5. The method of claim 2 wherein the encoding sentence embedding is followed by, prior to the grouping, performing dimensionality reduction on the encoded sentence embeddings.
6. The method of claim 5 wherein dimensionality reduction comprises performing Uniform Manifold Approximation and Projection (UMAP).
7. The method of claim 5 wherein dimensionality reduction comprises performing t-Distributed Stochastic Neighbor Embedding (t-SNE).
8. The method of claim 1 wherein the grouping comprises performing clustering.
9. The method of claim 8 wherein the clustering comprises performing Kmeans clustering.
10. The method of claim 8 wherein the clustering comprises performing Concensus clustering.
11. The method of claim 1 wherein the selecting is preceded by performing cluster splitting.
12. The method of claim 11 wherein the cluster splitting comprises performing splitting clusters into clusters for label generation and clusters for label ranking.
13. The method of claim 1 wherein the creating of the set of candidate intent names comprises performing label generation.
14. The method of claim 13 wherein the label generation comprises performing Generative Pre-trained Transformer 2 (GPT-2).
15. The method of claim 13 wherein the label generation comprises performing Bidirectional Encoder Representations from Transformers (BERT).
16. The method of claim 13 wherein the creating of the set of candidate intent names comprises performing simplification on the generated labels.
17. The method of claim 1 wherein the step of rating comprises ranking for most likely to least likely using a statistical model trained using a dataset for semantic similarity matching of labels to full sentences.
20. The computer program product of claim 18 wherein the grouping is preceded by encoding sentence embeddings contained in the corpus.
21. The computer program product of claim 20 where the encoding sentence embeddings comprises at least one process selected from the group consisting of a Language-Agnostic Bidirectional Encoder Representations from Transformers Sentence Encoding (LABSE) process, and a Robustly Optimized Bidirectional Encoder Representations from Transformers Pretraining Approach (RoBERTa) process.
22. The computer program product of claim 20 wherein the encoding sentence embedding is followed by, prior to the grouping, performing dimensionality reduction on the encoded sentence embeddings.
23. The computer program product of claim 22 wherein the dimensionality reduction comprises performing at least one process selected from the group consisting of a Uniform Manifold Approximation and Projection (UMAP) process, and a t-Distributed Stochastic Neighbor Embedding (t-SNE) process.
24. The computer program product of claim 18 wherein the grouping comprises performing at least one process selected from the group consisting of a Kmeans clustering process, a Concensus clustering process, and a cluster splitting process.
25. The computer program product of claim 18 wherein the creating of the set of candidate intent names comprises performing label generation.
26. The computer program product of claim 25 wherein the label generation comprises performing at least one process selected from the group consisting of a Generative Pre-trained Transformer 2 (GPT-2) process, a Bidirectional Encoder Representations from Transformers (BERT) process, and a simplification process.
27. The computer program product of claim 18 wherein the rating comprises ranking for most likely to least likely using a statistical model trained using a dataset for semantic similarity matching of labels to full sentences.
28. The system of claim 19 wherein the grouping is preceded by encoding sentence embeddings contained in the corpus.
29. The system of claim 28 where the encoding sentence embeddings comprises at least one process selected from the group consisting of a Language-Agnostic Bidirectional Encoder Representations from Transformers Sentence Encoding (LABSE) process, and a Robustly Optimized Bidirectional Encoder Representations from Transformers Pretraining Approach (RoBERTa) process.
30. The system of claim 28 wherein the encoding sentence embedding is followed by, prior to the grouping, performing dimensionality reduction on the encoded sentence embeddings.
31. The system of claim 30 wherein the dimensionality reduction comprises performing at least one process selected from the group consisting of a Uniform Manifold Approximation and Projection (UMAP) process, and a t-Distributed Stochastic Neighbor Embedding (t-SNE) process.
32. The system of claim 19 wherein the grouping comprises performing at least one process selected from the group consisting of a Kmeans clustering process, a Concensus clustering process, and a cluster splitting process.
33. The system of claim 19 wherein the creating of the set of candidate intent names comprises performing label generation.
34. The system of claim 33 wherein the label generation comprises performing at least one process selected from the group consisting of a Generative Pre-trained Transformer 2 (GPT-2) process, a Bidirectional Encoder Representations from Transformers (BERT) process, and a simplification process.
35. The system of claim 19 wherein the rating comprises ranking for most likely to least likely using a statistical model trained using a dataset for semantic similarity matching of labels to full sentences.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
December 16, 2020
November 22, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.